The growing deployment of video-surveillance systems is a critical component of the Department of Homeland Security's mission to secure the United States and its citizens from terrorist threats or attacks. As these systems gain in complexity, both in scope and feature sets, the need for intelligent video-content-analysis (VCA) technology becomes more apparent.
Intelligent video--using advanced technology to detect, classify, track, and analyze video information--enables video surveillance to shift from forensic to preventive systems. By identifying potential threats, such as unauthorized personnel entering a restricted room or a suspicious package left in a public area, surveillance systems that use VCA technology are more efficient and cheaper to operate than passive video systems.
However, adding intelligent video capability to a surveillance system means that the system must have additional, often significant, processing capability. In addition to handling the usual video-processing algorithms (compression, networking, and other control functions), the processors must now deal, in real time, with the algorithms that give "intelligence" to VCA-enabled systems. This requires special processor architectures.
Multicore processor platforms--comprising both DSP cores for data processing and general-purpose processor (GPP) cores for network and other control functions--represent a type of architecture that's well suited for intelligent-video surveillance applications. Along with the physical multicore silicon, there needs to be a flexible and easy-to-program development system that goes with the chip. Therefore, system integrators will be able to embed the often proprietary intelligent video algorithms on the platform.
As security concerns escalate, the need for and deployment of intelligent video surveillance systems will continue to increase. At the heart of these systems, multicore programmable processing will emerge as a very viable technology for successfully deploying VCA in a broad range of surveillance applications.